A site-wide SEO audit can take days. A single page takes about a minute, and most of what search engines judge happens at the page level: the title they show in results, the heading they read first, whether the page is even allowed in the index. The checker above runs those checks for you. Here is what each one means and what to change when it flags something.
Work one page at a time, and start with the pages that already get impressions in Search Console. A page sitting on the edge of the top ten is where a small fix pays back fastest.
What a page-level SEO check looks at
Every check reads the HTML your server sends. That is the version a crawler sees before any JavaScript runs, so it is the version that decides how the page is indexed.
The report groups the checks the way you would fix them. On-page tags and content come first, then whether the page can be crawled and indexed, then links, then the extras that help in specific places, and page weight last. A red item is worth stopping for. An amber item is worth a note. A green item needs nothing from you.
Title tag and meta description
The title tag is the line Google shows as the clickable headline. It is the single tag most likely to move both rankings and clicks, so it is the first thing to get right.
Write it for the person scanning a results page. Put the main term near the front, keep it to roughly 50 to 60 characters so Google shows the whole thing, and make it read like a sentence a human wrote rather than a list of keywords.
The meta description does not affect rankings, but Google often uses it as the grey snippet under the title. A description that answers "what will I get if I click" lifts the click-through rate. Leave it blank and Google writes its own from the page text, which is usually worse.
| Tag | What good looks like | Common problem |
|---|---|---|
| Title | 50 to 60 characters, main term first, one per page | Truncated, duplicated across pages, or missing |
| Meta description | 70 to 155 characters, written as a promise | Empty, or the same text on every page |
| H1 | One per page, close to the title in meaning | Missing, or several H1s from a page builder |
One H1, then a heading order that makes sense
The H1 is the page's on-screen headline. Use one. It should say roughly the same thing as the title tag, in slightly more natural language.
Below it, the H2s and H3s are the outline of the page. Search engines use that outline to work out what the page covers and to pull sections into featured snippets. Keep the levels in order: an H2 followed by an H3, never an H2 that jumps straight to an H4. The checker prints the full outline so you can read it like a table of contents and spot where it breaks.
A quick test: if the first sentence under each heading does not answer that heading on its own, the structure needs work before the wording does.
Enough content, and your term used the way a person would
Thin pages rarely rank for anything competitive. There is no magic word count, but a page trying to rank for a real query usually needs a few hundred words that answer it properly.
The keyword density table catches the opposite problem. If one word or phrase is a large share of the text, the page reads as written for a robot, and current ranking systems treat that as a negative signal. Your main term appearing naturally a handful of times is fine. Your main term at five percent of every sentence is not. Write for the reader and the density takes care of itself.
The checker reads text from the main content area only, with navigation, scripts and code stripped out, so the density numbers reflect what a reader sees rather than what is buried in the page source.
Canonical tags and staying in the index
A canonical tag tells Google which URL is the real one when the same content is reachable at more than one address. The safe default is a self-referencing canonical, where the page points at itself. The checker flags a canonical that points somewhere else, because that is how pages get dropped from the index by accident.
Two other things in this group are quiet page killers. A noindex robots tag removes the page from results, and it is easy to leave one on after a staging site goes live. A missing viewport tag means the page is not treated as mobile friendly. Both show up red for a reason.
The checker also fetches your robots.txt and looks for an XML sitemap, since those decide whether search engines can reach the page at all.
The links on the page
Internal links are how search engines find your other pages and how ranking strength moves around your site. A page with two or three internal links is hard to discover and hard to rank. Link to the related pages you already have, with anchor text that describes where the link goes.
The report lists every internal and external link with its anchor text, so you can see at a glance whether the important links use real words or say "click here". It also flags links with no text at all, which are usually an icon or an image that never got a label.
Structured data
Schema markup is a block of JSON in the page that tells search engines what the page is, in a format they do not have to guess at. It is what makes a page eligible for review stars, FAQ drop-downs, recipe cards and the other rich results, and it feeds the summaries that AI answers are built from.
The checker parses the JSON, lists which types are present, and shows the exact error if a block will not parse. Invalid JSON is skipped entirely by Google, so a broken Article block does nothing even though it looks fine in the page source.
The share preview
Open Graph tags control the card that appears when someone shares your link on LinkedIn, Facebook, Slack or in a message. X uses its own card tags, which fall back to the Open Graph ones. Without them, a shared link is a bare URL with no image and no description, and it gets far fewer clicks.
The report renders the preview card the way those platforms will, using your real title, description and image, so you can catch a missing image or a cut-off description before you publish.
Page weight and the speed you can see in the HTML
A full lab performance score needs a tool that loads the page in a real browser. What the checker can measure without one is the structure that sets the ceiling on how fast the page can be.
It counts the size of the HTML document, the number of separate script and stylesheet files, how many scripts sit in the head with nothing telling the browser to wait, how many requests go to other domains, and whether images below the fold are set to load lazily. A page with eight render-blocking scripts and a 300 kilobyte HTML file will be slow whatever else you do, and these are the numbers that tell you where to start.
HM Towhidul Islam · Senior Technical SEO Lead
I run the check on a page, fix only the red items, then run it again before I touch anything amber. Half the amber warnings clear themselves once the real problems are gone, and you avoid changing five things at once and not knowing which one helped.
Local signals, if you serve a place
Skip this group if your business is not tied to a location. If it is, the checker looks for LocalBusiness schema with your name, address, phone number and opening hours, and for a clickable phone number and address on the page itself. Those are the details Google needs to build a full business panel and to match the page to "near me" searches.
Turn the report into three or four fixes
A long report is easy to ignore. The summary at the top ranks every problem by how much it costs you, so the practical move is to take the first three or four items, fix those, and run the check again. That is usually enough to move a page that was close.
The items that tend to matter most, in order: a missing or weak title, a page set to noindex by mistake, no H1 or a broken heading order, thin content on a page targeting a real query, and too few internal links. Everything below that is polish.
Illustrative example
A service page stuck on page two
Picture a page that ranks around position 14 for its main term. The check finds a title that repeats the site name twice and runs past 70 characters, a single paragraph of body copy, and two internal links, both from the footer.
Rewriting the title to lead with the service and the city, expanding the page to answer the three questions buyers ask, and adding four in-content links from related pages is the kind of change that moves a page like this onto the first page over a month or two, with no new backlinks involved.
What to check on every page before you publish
- One title tag, 50 to 60 characters, main term first
- A meta description written as a reason to click
- Exactly one H1, and a heading order with no skipped levels
- A self-referencing canonical, and no stray
noindex - Enough body copy to answer the query, with the main term used naturally
- Four or more internal links with descriptive anchor text
- Valid structured data for what the page is
- Open Graph and X card tags with a working image
- Images below the fold set to load lazily
Run the check at the top of this page whenever you publish or make a big edit, then again a week later once Google has re-crawled.